Efficiency Evaluation of Brazilian Electrical Distributors Using DEA Game and Cluster Analysis

This paper presents a methodology for benchmarking identification in DEA combining DEA Game cross efficiency model and cluster analysis in a sequential way. The DEA Game seeks, in a non-cooperative game, not just the ideal efficiency for one decision making unit (DMU), but also for all the others. Since the traditional approach of DEA Game only provides efficiency rates this study will also make a cluster analysis using hierarchical clustering technique, known Ward's method, in order to obtain realistic benchmarks. These benchmarks will be obtained in the cluster analysis, where similar units will be grouped, by defining as benchmark the DMU more efficient in each cluster. Allowing in this way the inefficient units set tangible goals and objectives to improve their performance in the future. As an application of the proposed methodology, 61 Brazilian electrical distributors are considered.